23
$\begingroup$

$\DeclareMathOperator\B{B}\DeclareMathOperator\SU{SU}$Given a principal $E_8$ bundle $P\rightarrow X$ one can take the adjoint representation $\rho :E_8\rightarrow \SU(\mathbb C^{248})$ and form the associated vector bundle $V=P\times_{\rho}\mathbb C^{248}$. This vector bundle has an $E_8$ characteristic class $\lambda (V)\in H^4(X,\mathbb Z)$ obtained by pulling back the generator $\lambda \in H^4({\B}E_8,\mathbb Z)$ and it has a second Chern class $c_2(V)$ given by pulling back the generator $c_2\in H^4({\B}{\SU}(248),\mathbb Z)$.

I am looking for a reference for the following fact: $$\lambda (V)=\frac{c_2(V)}{60}.$$

There is the map ${\B}\rho^* : H^4({\B}{\SU}(248),\mathbb Z)\rightarrow H^4({\B}E_8, \mathbb Z)$. Since both groups are canonically isomorphic to $\mathbb Z$, the map is determined by a single integer, which is the Dynkin index of $E_8$ and has been computed to be $60$. The above fact essentially follows from this. I am writing a paper and would prefer to just state the fact and then point the reader to whoever first presented a thorough argument filling in all the details. I have seen it mentioned as a footnote on page 68 in Vector Bundles And F Theory by Friedman, Morgan, and Witten and some of the details regarding the Dynkin index are discussed in The torsion index of $E_8$ and other groups by Totaro.

$\endgroup$
1
  • $\begingroup$ I don't have an exact statement of this but I think it follows pretty simply from pages 452-454 of MF Atiyah, NJ Hitchin and IM Singer. {\sl Self-duality in four-dimensional Riemannian geometry}, Proc Roy Soc London Ser A, 362(1711), 425-461, 1978 and the fact that the first Pontrjagin class is \frac{1}{2} the second Chern class of the complexified tangent bundle. $\endgroup$ Commented Nov 12, 2016 at 12:30

0

You must log in to answer this question.